Acute Kidney Injury Requiring Renal Replacement Therapy, Kidney Replacement Therapy, Dialysis, Acute Renal Failure

Table III.

Oliguria	Urine output <200 ml/12 hours
Anuria	Urine output 0-50 ml/12 hours
Uremia	[Urea] > 35 mmol/l
Uremic Complications	Encephalopathy, pericarditis, myopathy
Fluid overload	Unresponsive to diuretics
Hyperkalemia	[K+] >6.5 mmol/l or rapid increase
Hyper/hyponatremia	[Na+] > 160 or < 110 mmol/l
Severe metabolic acidosis	pH < 7.1
Hyperpyrexia	Temperature > 40°C
Hypercreatininemia	[Creatinine] > 400 mcg/l
Overdose with dialysable toxin	Lithium, etc.

If one criterion is present then RRT should be considered; if more than one criteria are present then it is strongly recommended.

Oliguria and furosemide

In established olgiuric renal failure, fluid overload can cause a myriad of clinical sequelae, and there is emerging clinical evidence that excessively positive fluid balances are associated with worse patient outcomes. Patients with oliguric renal failure have significantly worse outcomes than those who maintain a urine output.

A subset of patients, when administered the loop diuretic furosemide, may convert from oliguric renal failure to non-oliguric renal failure. It is unclear if furosemide terminates the natural physiological response of salt and water retention in the face of the multiple insults associated with critical illness, a response that may become pathological if allowed to persist, or if this subset of patients suffer from a less severe form of AKI.

What is clear is that the administration of loop diuretic in the setting of oliguric AKI is commonplace. A multinational survey of primarily academic institutions demonstrated that bolused furosemide, titrated to the physiological end-point of urine output, is commonly used in the critically ill population with AKI, most often when AKI is associated with pulmonary edema.

As regards clinical evidence of benefit, post-hoc analysis of the BEST study showed no increase in mortality with the administration of furosemide in over 1700 critically ill patients with AKI. A meta-analysis of randomized controlled trials of the use of loop diuretic in renal failure revealed only five trials of suitable methodological quality for inclusion. Furosemide administration improved urine output and reduced the duration of RRT, but did not reduce mortality or increase RRT independence.

The significant difference between practice and the current evidence suggests clinical equipoise, and the need for high-quality evidence in this area. The SPARK study is a double-blind randomized controlled trial designed to address this issue. At present a furosemide challenge in bolus or infusion form may be given, but the aim should be the promotion of diuresis, rather than renal support or outcome modification.

Timing of RRT

At present the initiation of RRT is not an evidence-based process. The lack of consensus definitions for AKI has, until recently, prevented high-quality synthesis of the trials produced to date, as has variation in the criteria for RRT initiation. Oliguria, acidosis, increasing age and increasing number of organ failures are all strongly associated with an increased mortality risk, and it may be that this population may benefit from early intervention.

A large observational study has shown that increasing length of time from ICU admission to initiation of RRT is associated with an increased risk of mortality. There is no reproducible evidence for altered outcome based on initiation as a result of serum creatinine, urea or other variable.

Clinical decision-making: RRT and RST

At present consensus definitions and large clinical trials are required to assess the impact of timing of RRT on outcome in AKI. The opinion-based clinical algorithm in
Figure 4 was recently published to help with clinical decision-making regarding the intitiation of RRT and RST.

Figure 4.

How do I know this is what the patient has?

Rising serum and creatinine concentrations in the presence of the typical electolyte disturbances found in AKI, with or without oligoanuria, are the key features of AKI representing the loss of filtration and excretory function. Some drugs, notably trimethoprim, can cause elevations in serum creatinine, and urea is elevated in any catabolic state and by the use of steroids.

Urethral obstruction and trigone infiltration causing bilateral ureteric obstruction are potentially ameliorable causes of anuria and AKI. Obviously urinary catheter obstruction should be considered. Bladder scanning and abdominal ultrasound are useful bedside tools for assessing for obstruction of the renal tract.

4. Specific Treatment

Technique

The different forms of RRT vary in intensity, duration, frequency, mechanism of transport, type of filter, vascular access and infrastructure requirements.

IHD has already been described. The theoretical advantages of intermittent techniques are the potential to rapidly correct volume and electrolyte status, reduced monetary cost and reduced duration of specialist/intensive nursing input. The disadvantages are inferior solute and electrolyte control, fluctuations in intracranial pressure secondary to fluid shifts with rapid solute exchange, and increased hemodynamic instability. It should not be used in cases of cerebral edema or fulminant liver failure.

IHD also requires larger catheters and higher blood flows than CRRT to be effective. IHD may also require heavy involvement of the local renal service, which may not be feasible due to service constraints.

The theoretical advantages of CRRT are that they should allow continuous and tightly regulated control of the internal milieu, even in the hemodynamically unstable critically ill patient. Many of the complications of CRRT have been ameliorated by the adoption of veno-venous techniques over arteriovenous methods. It is labor- and cost-intensive. Clotting of the filter results in discontinuous therapy, a loss of fluid and hemoglobin and a reduction in delivered dose.

Despite the theoretical advantages of CRRT, multiple systematic reviews and meta-analyses have failed to demonstrate a mortality benefit in using one technique over the other. Biased selection of patients, with more unwell patients receiving CRRT, may partially explain this finding.

However, the studies have been performed over a 20-year period. Over the past 10 years RRT technologies have evolved into substantially different therapies from their predecessors. The patient demographic admitted to ICU and commenced on RRT is also significantly different, and this may be a more powerful reason for the lack of demonstrated mortality benefit in meta-analysis.

CVVH or CVVHDF?

A direct clinical comparison between CVVHDF and (pre-dilution) CVVH has shown that CVVHDF is superior at clearing small molecules like urea, but the techniques have equivalent clearances for middle molecules, represented experimentally by beta2-microglobulin, at blood flow rates of 125 ml/min. Theoretically one would assume the purely convective method to be superior at clearing middle molecules, and this is likely a function of pre-dilution. The effect of higher blood flows on solute clearances has not yet been demonstrated.

CVVHDF is, at modest flow rates, the most efficent choice of modality to achieve broad solute clearance. Optimal function is likely to be dictated by a specific set of operative conditions, as described below in “Prescribing CRRT.”

Dose of CRRT

RRT is a complex intervention, and assessing the efficacy of a modality can be difficult. The most pragmatic method of estimation is the intensity or efficacy of clearance of a surrogate solute, commonly urea, though this is imperfect for reasons outlined elsewhere. There is no evidence to say that control of measured solute is more important than control of tonicity, volume or acid-base status.

There have been numerous studies looking at the effect of delivered dose on renal and patient outcome in the ICU. These have mainly been single-center studies, and of varying methodological quality, using a variety of RRT modalities and recruiting only several hundred patients.

The ATN trial in 2008 and the RENAL trial in 2009 were large, multi-center, randomized controlled trials, each recruiting over 1000 patients. They were designed to specifically examine the effect of delivered dose on patient outcome. Neither trial showed a significant difference in either renal recovery or mortality between the high- and low-dose groups.

A large prospective, multi-center observational study confirmed these results. Three recent meta-analyses reviewing the data for over 3000 patients agree that, although the studies show significant heterogeneity, doses of RRT >25 ml/kg/hr or the equivalent do not reduce mortality or aid renal recovery.

It is unlikely that there is no dose-dependent difference in outcome. It may be that 25 ml/kg/hr is the optimal dose for RRT. It is more likely that we are not considering the correct variables and that timing, modality or other unresolved issues have a significant opposing impact on outcome.

Clinical dosing is another matter. The dose of RRT required to correct dangerous hyperkalemia in one patient will be significantly different from that required for managing fluid overload with hemodynamic instability in another and needs to be tailored to the clinical situation.

Anticoagulation

Essentially, critical illness is a prothrombotic state. Uremic platelet dysfunction can result in bleeding. The components of the ECC in patients undergoing RRT can be prothrombotic, and kinks in the circuit or problems with vascular access can lead to blood stasis. Clotting of the filter or circuit is wasteful, contributes to anemia, and interrupts the delivery of RRT.

A variety of techniques can be used to extend filter life. Simple techniques like post-dilutional transfusion of blood, the use of pre-dilutional fluid replacement, ensuring vascular access is adequate and maintaining blood flows of 200 ml/hr or more have been shown to extend filter life. These are particularly useful in patients with variceal bleeding, recent surgery or intracranial hemorrhage.

Anticoagulation can be local or systemic. Local anticoagulation involves predilutional administration of an anticoagulant with post-dilutional administration of an antidote. This may be heparin/protamine or citrate/calcium. It is more complex than systemic anticoagulation, but so long as the anticoagulant is neutralized, it should be safe in those patients for whom systemic anticoagulation is contraindicated.

Systemic anticoagulation is normally by unfractionated heparin to allow easy monitoring, reversal and titration. Low-molecular-weight heparin is more complex to monitor and needs to be adjusted depending on the effective clearance. They may both cause heparin-induced thrombocytopenia, though it should be noted that there are other causes of thrombocytopenia in patients on CRRT. Heparinoids or prostacycline can be used as alternatives.

Filter clotting should not prompt an automatic escalation in anticoagulation; the etiology must be investigated. Larger-diameter vascular access catheters, reducing the ultrafiltration rate, or increasing blood flows may be more effective, without increasing the risk of hemorrhage.

CRTT in specific circumstances

Hemodynamic or neurological deterioration, other organ dysfunction, impaired endogenous clearance or clinical deterioration despite best supportive care in cases of poisoning may prompt the use of RST in the form of hemodialysis or hemofiltration.

The topic of blood purification for poisoning is not reviewed in detail here. CVVHDF is likely to be the most successful at drug removal. For heavily protein-bound drugs like theophylline and barbiturates, hemoperfusion, an absorptive modality of solute separation, using a charcoal filter may provide superior drug clearance.

Therapeutic plasma exchange, using an ECC much like RRT with the addition of a pheresis machine to separate and return cellular elements to the patient while discarding and replacing the toxin-containing plasma, is occasionally used for the treatment of poisoning where the substance is question is of very high molecular weight or is extremely protein-bound. Examples of such toxins include cisplatin and snake venom.

Immunomodulation

There is a physiological premise supporting the use of specialized convective forms of CRRT in the removal of cytokines and related inflammatory mediators in the middle molecular range. The underlying pathophysiology is too complex for a worthwhile examination here, but Phase I and II studies in humans are ongoing after initial animal models suggested that such intervention may attenuate the inflammatory response responsible for the sepsis syndrome. It is likely to be some time before large-scale, suitably powered, randomized controlled trials are available to assess the impact of such therapies on outcome.

Liver failure

The syndrome of liver failure involves the accumulation of heavily protein-bound toxic substances with profound neurological and systemic effects. CVVHDF is effective at removing water-soluble toxins and may be useful in controlling fluid balance and preventing the accumulation of the uremic toxins, but additional mechanisms of solute removal are required to effectively provide support for the failing liver.

Several techniques of extracorporeal hepatic support have been developed, of which the molecular adsorbents recirculating system (MARS) is best known. Adequately powered clinical trials are awaited: to date there has been no impact on outcome, though they have been shown to be safe and effective at improving biochemistry.

Prescribing renal replacement therapy

RRT is no different from other interventions in critical care: it requires operational parameters to be set. The operational parameters required to achieved a desired efficacy, or dose, of CRRT should be thought of as a prescription with the same requirements for precision and accuracy.

Technique/mode

There will often be only a single mode of RRT offered by a unit. In central units with the infrastructure to offer multiple modalities the desired mode needs to be clearly stated. This will affect both circuit design and set-up.

Dose

Effective dose depends on clearance. Traditionally it is described in ml/kg/hr. Most centers aim for 25-35 ml/kg/hr. This should be stated, and the desired dose will determine the other operational parameters depending on the modality selected.

Fluid balance

Goals should be set for the patient – for example, -1000 ml over the 12 hours from 1200 to 0000 – and for the machine. The administration of drugs needs to be taken into account and the machine goals – for example, -200 ml/h – adjusted to meet the patient goals. Fluid can be administered if a positive balance is sought by increasing replacement fluid or reducing ultrafiltration.

Problems can arise with failure to meet targets due to significant downtime with filter crashes or multiple procedures or investigations requiring cessation of RRT. Incorrect machine settings or periods of low blood flow may also contribute, as may hemodynamic instability precluding fluid removal.

Flow rates

Blood flow, Effluent flow, Dialysis flow and Replacement fluid flow should all be prescribed in accordance to dosing and fluid balance goals.

Anticoagulation

The anticoagulant infusion rate, concentration and the drug itself need to be prescribed clearly, and orders need to be given for the antidote in regional anticoagulation.

Drug dosing in RRT

Changes in drug handling

Severe AKI and its consequences have profound effects on pharmacodynamics and pharmacokinetics. Hypoalbuminemia increases the free fraction of protein-bound drugs available for clearance AKI and critical illness are dynamic process, and the clearance of drugs can vary significantly from day to day. Drug removal by RRT depends on residual renal function, the mode and technique of RRT, the principle of solute removal adopted, and the physicochemical and pharmacokinetic properties of the individual drug.

CVVHDF is likely to be the most effective method of drug clearance. Drugs of low molecular weight, with low volumes of distribution, poor protein binding and high renal clearance, are most effectively cleared by CRRT. Specific dosing information is outside the realms of this work. Prescribing for patients on CRRT requires the support of experienced ICU pharmacists, and texts such as the Renal Drug Handbook, the British National Formulary or other sources should be consulted.

5. Disease monitoring, follow-up and disposition

Response to CRRT

The clearance of uremic toxins can be monitored using urea and creatinine as surrogates. This is imperfect, as discussed elsewhere in this chapter. Patient weight or fluid balance charting will show the efficacy of ultrafiltration in the management of fluid overload. Acid-base status, electrolyte abnormalities, and uremic markers should stabilize after 24-48 hours.

Improvement in the clinical condition of the patient – resolution of uremic encephalopathy, improved hemodynamic stability, improvement in the organ systems for which RST was commenced – is one of the most important methods of monitoring response to therapy.

Metabolic monitoring

Replacement or dialysis fluid contains sodium, chloride, lactate or bicarbonate as a buffer, and calcium and can contain varying concentrations of potassium in order to promote physiological homeostasis as part of RRT. However, after prolonged RRT, magnesium, phosphate and other trace elements may be depleted. Daily monitoring of electrolytes is mandatory, and replacement of phosphate and magnesium will likely be required.

Monitoring anticoagulation

APTT should be checked regularly if using heparin-based anticoagulation, be it local or systemic. Protamine or calcium administration rates can be increased if systemic coagulation is compromised. Local protocols should exist to guide therapeutic anticoagulation to local laboratory standards, particularly if specialist techniques or low-molecular-weight heparins are employed. Platelets should also be monitored.

Ineffective RRT

If the physician and nursing staff prescribing and operating the CRRT system have a sound understanding of the circuit, the modalities, the method of fluid replacement and the choice of anticoagulation, then reliable control of uremia and fluid balance with an adequate filter life and minimal technical issues can be safely and effectively achieved.

Ongoing training and a supportive infrastructure are mandatory in the provision of RRT for 24 hours a day. Rapid recognition of signs heralding machine dysfunction or patient deterioration can ameliorate many problems experienced while running CRRT.

Frequent filter clotting is likely to be the biggest contributor to ineffective RRT. Inadequate vascular access limits blood flows and can limit the effectiveness of solute clearance. It can also lead to reduced filter lifespan, as can sub-therapeutic anticoagulation, high rates of ultrafiltration and post-dilutional fluid replacement. Increasing transmembrane pressure can herald incipient filter failure. Recognition can allow the return of the blood in the ECC to the patient to minimize blood loss.

Cessation of renal replacement therapy

Renal replacement and support

If CRRT has been started as a replacement therapy, for control of electrolyte or fluid disturbance, or for treatment of the uremic syndrome in the presence of AKI, then adequate control of these variables needs to be achieved prior to cessation. The main mechanism of this will be recovery of functional urine output – i.e. recovery of renal excretory function.

Increasing urine output and falling creatinine concentration in the context of a fixed CRRT prescription and a clinically stable patient suggests recovery of excretory function. CRRT commenced in support of other organ dysfunction should be ceased once there is objective evidence of organ recovery.

There is a marked paucity of evidence on this topic. Practices vary significantly. Large clinical trials are badly needed to improve patient care.

Withdrawal of care

In some patients the progression to AKI requiring RRT is used as a ceiling of therapy by physicians. Ongoing deterioration despite maximal therapy to that point is taken as a sign of futility and, depending on the age, predicted mortality, the likelihood of functional recovery, the wishes of the family and the premorbid wishes of the patient, CRRT may not be started. In some cases where expected mortality is very high a trial of therapy over a clearly defined time frame may be indicated.

Where treatment has already begun but clinical deterioration has continued, RRT may be withdrawn along with vasoactive and ventilatory support. This is generally done on medical grounds with the consent of the family and CRRT is stopped.

Monitoring treatment cessation

It should be expected that urea and creatinine will rise after cessation of CRRT. The recovering renal function is unlikely to be as great as that of residual renal function and RRT clearance. They should plateau within 48 hours; continued rise after this time may indicate that CRRT needs to be re-established. The trajectory of the rise should also be considered, as should whether the volume of urine being produced is sufficient to maintain homeostasis.

As mentioned above, furosemide is commonly used to attempt to provoke a diuresis in patients with AKI. A significant number of physicians use furosemide in the recovery phase of AKI post-RRT in an attempt to increase urine output to accelerate weaning from CRRT, though there is no evidence to support this.

Post-hoc analysis of the BEST Kidney study showed that increasing urine output pre-cessation and falling creatinine levels pre-cessation are the best indicators of successful weaning from RRT, though the predictive ability of urine output was adversely affected by the use of diuretics. Early oliguria, prolonged RRT, increasing age and increasing illness severity scoring have been associated with failure to wean from CRRT.

Moving to chronic RRT

There is very little evidence regarding the conversion of CRRT for AKI to IHD once the only requirement for ICU management is non-recovery of renal function. Dialysis dose has been treated as a fixed quantity in the literature. There are no large trials demonstrating either benefit or harm in converting to IHD under these circumstances, even with a significant reduction in dialysis dose.

Most units will wait until the clinical condition has stabilized and reliable control has been achieved over the internal milieu before converting to IHD. There is no evidence to guide practice on this topic, and local practice varies considerably.

Follow-up

Patients with any form of AKI are at substantially higher risk of in-hospital mortality. If there is an ongoing need at discharge from the ICU for further investigation as to the cause of AKI or ongoing RRT is required in the form of IHD for non-recovery of renal function, then the local renal team should become involved. It may be that this requires transfer to another hospital. Even if excretory renal function is recovered, then patients should still be discharged to the care of physicians with acute care experience.

Pathophysiology

Pathophysiological consequences of severe AKI

Resolution of severe AKI can take days, weeks or even months, during which time the loss of filtration function, the retention of fluid and waste products and the loss of the normal mechanisms of hormonal and metabolic homeostasis can lead to profound and complex pathophysiological consequences. Known collectively as the uremic syndrome, this derangement of biochemistry and normal physiology parallels the severity of the AKI and has profound systemic effects.

Cardiovascular

Loss of urine output leads to fluid retention and, in combination with hypoalbuminemia, may lead to congestive cardiac failure. AKI can lead to a SIRS response with the potential for adverse hemodynamic effects. Volume overload can also result in secondary hypertension.

Respiratory

Volume overload and hypoalbuminemia lead to pulmonary edema and pleural effusions. Reno-pulmonary crosstalk can potentiate acute respiratory distress syndrome (ARDS) in those patients at risk with infiltration of the lung tissue by activated cytokines. Uremic inhibition of platelet function can increase the risk of pulmonary hemorrhage.

Immunological

A number of molecules normally excreted in the urine but retained in AKI have the potential to modify the immune response. Fluid retention and hypoalbuminemia cause tissue edema. White cell dysfunction occurs with the promotion of free radical formation and oxidative stress. The net effect is of delayed tissue healing and increased susceptibility to sepsis.

Gastrointestinal

Volume overload and gut edema can lead to impaired barrier function, impaired absorption and abdominal compartment syndrome. The inflammatory and hemodynamic consequences of AKI can lead to gut ischemia and gastroduodenal ulceration.

Hematological

Loss of erythropoietin, increased erythrolysis and impaired erythropoiesis promote anemia. Bleeding may be a significant complication of AKI with uremic inhibition of platelet aggregation.

Neurological

Uremia, disorders of electrolyte homeostasis, malnutrition and altered drug handling can all result in, or propagate, obtundation, oversedation, ICU delirium, prolonged ventilation and myoneuropathy.

Acid-base and electrolyte homeostasis

This can be profoundly deranged with retention of acid, accumulation of organic acid, reduced chloride excretion, hyperkalemia, dilutional hyponatremia, reduced buffering and hypoalbuminemia. Malignant arrhythmias, neurological side effects, marked unantagonized catabolism, hemodynamic dysfunction and lung and gut injury may all occur as a consequence.

Drug handling

Both pharmacodynamics and pharmacokinetics are significantly altered in AKI. Fluid retention significantly increases the volume of distribution, while elimination, protein binding and availability may all be siginifcantly reduced depending on the drug in question. Drug toxicity can occur at much lower doses than in patients with intact excretory function, but conversely, some drugs may be underdosed.

Uremic toxins

Much of the current understanding of the pathophysiology of uremia is extrapolated from work performed in CKD. This is a relatively stable condition without the additional confounders of sepsis, multiple organ dysfunction, and the systemic inflammatory response, which often coexist in the critically ill patient with dynamic AKI.

Both conditions occur as a result of a toxic accumulation of uremic toxins, and so they are likely to share some pathological mechanisms. In addition, the pathophysiological effects of RRT itself can obscure the clinical picture.

Though known as the uremic syndrome, over 90 solutes have been shown to be retained in renal failure. More await identification. It is uncertain how important each of these substances is to the genesis of the syndrome, but it is likely that some or all contribute to the pathophysiological consequences of severe AKI. They can be divided into small and middle molecules on the basis of molecular weight.

It is generally accepted that urea would be toxic only at concentrations higher than those normally found in AKI. However, the clearance of urea is a convenient measure of the efficacy of RRT, despite the different clearances, kinetics and volumes of distribution of each retention product.

Pathophysiological consequences of RRT for severe AKI

Though RRT undoubtedly saves lives and is able to correct many of the adverse pathophysiological consequences of severe AKI and uremia, it may also cause adverse effects in its own right.

Hemodynamic disturbance

The extension of the circulation on initiation of CRRT can drop the hematocrit and may affect tissue oxygen delivery. It may also cause hypovolemia.

Microbiological contamination

Unless a reliable source of ultrapure dialysate or replacement fluid is available, then the risks of infection, pyogenic transfer and a further inflammatory insult are increased. In addition, manufacturing error or deliberate inoculation can lead to contamination of otherwise sterile commercial product.

Membrane compatibility

Older biologic membranes were associated with hemodynamic disturbance and the activation of complement and leukocytes. This lack of biological compatibility was thought to prejudice renal recovery; newer synthetic membranes do not seem to have the same disadvantage.

Vascular access

Inadvertant vascular damage may lead to anemia, hypovolemia and hemodynamic instability. As with any indwelling device, the risk of infection also increases.

Anticoagulation

Prothrombotic factors – inflammatory mediators, immobility, hypoperfusion – vie with anticoagulant factors – uremia, anticoagulants – in the critically ill patient. With AKI requiring RRT the natural clotting mechanisms are further compromised: the RRT circuit may be prothrombotic, while extending filter life to ensure continuous RRT often requires local or systemic anticoagulation.

Specific uremic retention products

Small molecule retention products

These are products with a molecular weight <500 Da. Asymmetric dimethylarginine accumulates, interfering with nitric oxide production and promoting endothelial dysfunction. Creatinine also accumulates in renal failure, though no evidence of toxicity has been demonstrated. Hippuric acid accumulation leads to altered drug protein binding and tubular transport of organic acids. It may push free, active drug concentrations into the toxic range. Indoxyl sulphate has similar effects. Hyperphosphatemia can lead to vascular damage through calcium-phosphate complex deposition. Other small molecules include purines, guanides and homocysteine.

Middle molecule retention products

These are those substances with a molecular weight between 500 and 12,000 Da retained in renal failure. Cytokines are the primary middle molecule to be retained. Septic patients with AKI can develop massive blood concentrations of cytokines, promoting the inflammatory response and subsequent organ dysfunction.

Other middle molecules include peptides that inhibit the activity and function of leukocytes, kappa and lambda light chains, themselves nephrotoxic, parathyroid hormone leading to intracellular calcium influx and beta2-microglobulin, also nephrotoxic and implicated in dialysis-related amyloidosis, though this is unlikely to be of significance in the acute setting.

Epidemiology

The problem with renal failure and AKI

The epidemiological study of AKI has been profoundly hampered by the lack of a consensus definition for the syndrome. Though many studies have been performed, their definitions of acute renal failure are so disparate that synthesis and meaningful review is very difficult. Very few studies to date have looked at early AKI or the importance of small increases in creatinine concentration. The introduction of the RIFLE classification system has provided a consensus upon which future studies can be based. Table IV lists the risk factors for the development of AKI.